Computer Assistance for Solving Imaging Problems

Joel A. Garcia, MDa,b,*, Babak Movassaghi, PhDc

KEYWORDS
 Computer assistance Angiography  Coronary  Rotational angiography  Foreshortening  Overlap  Optimal view map  Universal view map

Imaging artifacts inherent in two-dimensional (2D) projection of complex three-dimensional (3D) structures in standard angiography limit its usefulness in making vascular evaluations.1,2 Innovative 3D reconstruction software based on CT, magnetic resonance (MR), and planar imaging are major advances in the diagnosis and treatment of endovascular diseases. With the use of advance processing, these imaging techniques provide 3D vessel evaluations and characteristics that have important clinical implications. Meanwhile, standard angiography, while a well-established technique, has not been so readily used to provide 3D reconstructions because of imaging artifacts inherent in two-dimensional (2D) projection of complex 3D structures.1,2 Now, though, computers can be used to correct most of these known inaccuracies in x-ray based angiography.37 Furthermore, a computer-based technique applied in standard angiography can precisely characterize stent-induced conformational changes in three dimensions.812 In determining the risk associated with a given lesion, it is necessary to measure such vessel properties as tortuosity, length, and take-off angles, and to determine bifurcation and ostium characteristics.13 These properties and characteristics are best evaluated in 3D representations of the vascular tree. For this reason, images showing

dynamic conformational changes of the vascular tree should also play a role in the treatment planning process. Two-dimensional imaging techniques can also be used to represent and measure these variables. However, the accuracy of such representations and measurements is often limited.1,14 These 2D techniques of image acquisition with traditional angiography are nonstandardized, subjectively chosen, and are highly dependent upon the 3D visual skills of individual operators.4 By contrast, a 3D evaluation of a vessel yields an accurate representation of clinically relevant vessel properties and characteristics and incorporates important dynamic changes.8,12,1520 Threedimensional quantification of these changes eliminates dependence on the users visual estimation and standardizes the vessel-evaluation process, therefore minimizing inaccuracies. Contemporary medicine has evolved along with imaging techniques. The traditional 2D planar imaging evaluation is now complemented by biplane angiography, rotational angiography, CT, and MR. Some of these widely used techniques, such as CT and MR, already are used in making 3D vascular evaluations.2138 In an era of complex and expensive interventions with drug-eluting stents and other devices, precise length measurements and accurate

Cardiol Clin 27 (2009) 503512 doi:10.1016/j.ccl.2009.03.009 0733-8651/09/$ see front matter 2009 Elsevier Inc. All rights reserved.

cardiology.theclinics.com

a Medicine Department, Division of Cardiology, University of Colorado at Denver, 12401 E 17th Ave, Box B-132 Leprino Building, Rm 524, Aurora, CO 80045, USA b Medicine Department, Division of Cardiology, Denver Health Medical Center, 777 Bannock street, MC 0960, Denver, CO 80204, USA c Philips Healthcare, Research Department, 12401 E 17th Ave, Box B-132 Leprino Building, Rm 524, Aurora, CO 80045, USA * Corresponding author. Denver Health Medical Center (DHMC), University of Colorado Hospital at Denver and Health Sciences Center (UCDHSC), 777 Bannock Street, Mail Code 0960, Denver, CO 80204. E-mail address: joel.garcia@dhha.org (J.A. Garcia).

504

Garcia & Movassaghi

placement are critical to minimize the need for additional interventions resulting from inaccurate or incomplete imaging. Recognition of these limitations has resulted in the development of imaging techniques designed to specifically address the weaknesses of the traditional angiographic approach.8,9,16,19,3946 Advanced computer-assisted technologies capable of minimizing the shortcomings of traditional angiography have been developed and are now in clinical use. Three-dimensional reconstruction images can be generated from techniques that acquire volumetric data points. These techniques include rotational angiography, CT, and MRI. The surface-rendering method relies upon a computer algorithm to reconstruct intensity values. The resultant image is a representation of the surface contour and appears 3D through computer-generated shading. The maximum intensity projection algorithm is another commonly used surface-rendering technique. Volume rendering uses all the data and reconstructs all data points without shading or computer enhancement.

CURRENT ANGIOGRAPHIC TECHNIQUES

The placement of the gantry in a location to produce useful angiographic information is a fundamental task in both diagnostic imaging but also in the performance of endovascular interventions. Obtaining optimal angiographic views is critical to assessing lesion morphology, extent of disease, and involvement of major branch segments.14 These considerations have become more prominent since the advent of interventional cardiology because, with interventional cardiology, the precision is more important and evaluation goes beyond simply noting the quality of distal conduits for bypass surgery. Three-dimensional vascular trees registered or aligned in the coordinate system of gantry location can be used to perform the imaging tasks commonly encountered. First, overlap of vessels in the tree needs to be minimized. Second, segments of the tree need to be imaged with the imaging system perpendicular to the axis of the vessel segment such that no foreshortening is produced in the resultant projection image. Third, the specific bifurcation points need to be accurately imaged.

FORESHORTENING, OVERLAP, AND BIFURCATION LESIONS

Vessel foreshortening and overlap are recognized imaging artifacts that result from the 2D angiographic projection of 3D vascular structures.1,2 The tortuosity of coronary artery segments present specific challenges in minimizing vessel foreshortening and angiographically separating adjacent structures. The clinical implications of unrecognized foreshortening include missed lesions, errors in assessing lesion length, incomplete coverage of lesions by stents, underestimation of stenosis severity, and inaccurate quantitative coronary angiography calculations. Vessel overlap, a result of one vessel or segment superimposed on another, prevents complete image interpretation. As opposed to vessel foreshortening, vessel overlap is easily recognized by the operator and does not require significant image processing. Ostial lesions are traditionally difficult to image because of challenges in avoiding the ostium of the vessel to be covered by the main branch or a contiguous vessel. Screening angiographic evaluations are often limited in providing a complete ostial vessel survey, thus requiring multiple subsequent angiograms.

Coronary Modeling
The 3D modeling technique uses 3D centerline data and shaded or rendered surfaces; the diameter and 3D morphologic structure of the vessel is subsequently derived with a computer algorithm. This 3D modeling technique uses two or more angiographic projections to extract features of the vessel and create a 3D representation. Modeling can be obtained on line and off line from various imaging modalities. It can be obtained from standard angiography, rotational angiography, and extended acquisitions.

THE NEED FOR COMPUTER ASSISTANCE IN DETERMINING OPTIMAL VIEWS

Sometimes even experienced intervenionalists fail to choose the best view for minimizing image foreshortening of the diseased segment,14,41 even though several methods are readily available to produce useful images that avoid overlap and minimize foreshortening for all segments of interest in the vascular tree.3,3941 Computer graphics can be used to display the tree in a variety of views (3D modeling and 3D reconstruction) and the operator can select

Coronary Reconstruction
Several methods capable of generating 3D images have been described and, in general, can be classified as either surface-rendering techniques or volume-rendering techniques.

Computer Assistance to Solve Imaging Tasks

appropriate views (Fig. 1). Alternatively algorithms can be written to automatically process the data, recommend specific views, or produce visual guides that combine a parameter, such as the extent of foreshortening for a vessel segment of interest, for all possible angiographic views. The later approach is represented by a color-coded map of the degree of foreshortening and overlap for all possible gantry locations (Fig. 2). views is often costly in expending time, radiation, and contrast. Optimizing working views for interventions theoretically should reduce visualization-related mistakes and prevent complications. Several groups have demonstrated the value of using a 3D vascular tree to simulate angiographic views.5,6,4749

505

How is the Optimal View Map Created?

With the use of a rotational acquisition protocol (cranial or caudal), with any field of view, a 3D modeling of the left coronary artery (LCA) system is completed. The reconstruction is performed with the use of 3D modeling software commercially available via various vendors. Once the rotational run is completed, it is transferred to a dedicated 3D workstation where the analysis is performed. Current systems automatically transfer the image sequence to the workstation for modeling/reconstruction in under 60 seconds. The system is designed to generate a complete model of the artery after various steps. First, the operator must identify two orthogonal views of the target coronary tree. Second, both views should be completely or partially gated to the ECG. This step is not mandatory for modeling, but ensures greater precision. Third, points in the vascular tree must be recognized by clicking on them on both views. The generation of a computer-created 3D model follows. The results provide three important clinical results: (1) a 3D model that enables measurements of length and size, (2) an optimal view map, and (3) the gantry position of all intended views in the optimal view map (Fig. 3).

PATIENT-SPECIFIC OPTIMAL VIEW MAP

Three-dimensional vascular trees generated with CT angiography, MR angiography, or traditional x-ray technology can be used to simulate all possible angiographic views of the vascular tree. The principle that 3D data sets can be used to simulate 2D images is an important practical approach needed in endovascular interventions since the image-guidance of multiple techniques remains to be fluoroscopy. With the expanded use of CT angiography and MR angiography, making the best use of information from each diagnostic modality when the patient comes to interventional therapy is increasingly important. The clinical value of using a 3D vascular tree to simulate angiographic views is to enhance patient safety and potentially improve interventional outcomes. Computer computation of an optimal view can be done before the intervention as part of the planning process for the procedure. For the interventional procedures, the traditional trialand-error method of finding good angiographic

How is the Optimal View Map Used?

Using the modeled/reconstructed images obtained from the workstation, the best working view and lesion length can be identified with no extra contrast or fluoroscopy/cine use. The 3D-assisted best working view represents the least foreshortening. This avoids underestimation of the severity of the lesion, avoids treatment of side branches, and identifies the accurate length needed for stent selection (see Fig. 3). The current Phillips FD-20 system (Philips Healthcare, Best, The Netherlands), which generated the images for this article, is one of several imaging systems capable of automatically setting itself to the best working view.

Fig. 1. Three-dimensional modeling concept. Two orthogonal views result in the forward projection of a computer-generated image. (Adapted from Garcia JA, Movassaghi B, Casserly IP, et al. Determination of optimal viewing regions for x-ray coronary angiography based on a quantitative analysis of 3D reconstructed models. Int J Cardiovasc Imaging 2009;25:45562; with permission.)

UNIVERSAL OPTIMAL VIEW MAP

Using previously validated 3D modeling techniques in a large patient cohort undergoing coronary angiography, a recent study identified

506

Garcia & Movassaghi

Fig. 2. Generation of segment-specific optimal view map. (A and B) Two projection views of modeled left coronary artery corresponding to the overlap optimal view map (C) of the vessel segment shown as a green centerline in A and B. Projection image A corresponds to a yellow region and projection image B to a red region in the optimal view map (C) where yellow regions illustrate minimum vessel overlap and red regions illustrate severe vessel overlap. Similarly, 2d and 2e show two projection views of modeled left coronary artery corresponding to the overlap optimal view map (2f) of the vessel segment shown as a green centerline in 2d and 2e. CAUD, caudal; CRAN, cranial; LAO, left anterior oblique; RAO, right anterior oblique.

optimal view regions for first- and second-order coronary segments that minimized vessel foreshortening and overlap with the goal of reducing imaging inaccuracies. Within the viewing regions identified, the minimum vessel foreshortening was 5.8% 3.9% for the left coronary artery and

5.6% 3.6% for the right coronary artery, and the average overlap was 8.7% 7.9% for the left coronary artery and 4.6% 3.2% for the right coronary artery. This represents the first scientific validation of optimal viewing regions to guide diagnostic and interventional coronary procedures.41
Fig. 3. Three-dimensionalmodeling/reconstruction station. The coronary model allows for size and length evaluations, the optimal view map (color coded), and the gantry position. This example shows on the coronary model how the midleft anterior descending artery highlighted segment measures 22.3 mm. The optimal view map shows a white point (LAO 60 Cran 30 ) where there is 15.7% foreshortening. The black point represents a computer-suggested gantry with minimal foreshortening at 1.2%. Finally, the gantry shows its position at LAO 60 Cran 30 in the left lower corner. Caud, caudal; Cran, cranial; LAO, left anterior oblique; RAO, right anterior oblique.

Computer Assistance to Solve Imaging Tasks

In general, the optimal viewing regions for individual coronary segments defined in this study match viewing angles previously recommended by experts. However, there are some notable exceptions: the right anterior oblique cranial region proved optimal for viewing the midleft anterior descending artery, the posteroanterior caudal region for viewing the proximal left main, and right anterior oblique caudal region for viewing the mid posterior descending artery.41 For current interventional procedures, where decisions regarding device length may be hampered by failure to appreciate vessel foreshortening, the optimal viewing regions defined in the mentioned study may provide important scientifically based guidance that could be especially useful in laboratories where 3D modeling and patient-specific optimal view maps are currently unavailable. While we have sought to validate optimal viewing regions applicable to a broad population undergoing coronary angiography, we believe that the future will be dominated by a patient-specific approach that will deliver superior results (Fig. 4).

507

NEW IMAGING TRAJECTORIES

New x-ray systems are now capable of acquiring trajectories where the C-arm moves freely around the patient while capturing data. With the incorporation of the universal optimal view map, investigators are now able to design new imaging trajectories that can cover all required gantries for a complete diagnostic evaluation in a single sweep. These imaging trajectories aim at covering all spots in the view map that minimize foreshortening and overlap for each coronary segment (Fig. 5). A study is currently investigating the feasibility and clinical impact of these trajectories compared with standard angiography.

DATABASE CREATION
The generation of 3D vascular data on large numbers of patients undergoing diagnostic imaging with CT angiography, MR angiography, or 3D processed angiographic images provides an opportunity to catalog anatomical features in the form of databases. Subsequent evaluation

Fig. 4. Universal optimal view map. Combined universal optimal view map for the first-order vessel segments of the left coronary artery (A) and right coronary artery (RCA) (B). Combined universal optimal view map of the second-order vessel segments of the left coronary artery (C, D). Biff, bifurcation; CAUD, caudal; CRAN, cranial; d, distal; DIAG, diagonal; LAD, left anterior descending artery; LAO, left anterior oblique; LCX, left circumflex artery; LM, left main; m, mid; OM, obtuse marginal; p, proximal; PDA, posterior descending artery; PL, posterolateral; RAO, right anterior oblique. (Adapted from Garcia JA, Movassaghi B, Casserly IP, et al. Determination of optimal viewing regions for x-ray coronary angiography based on a quantitative analysis of 3D reconstructed models. Int J Cardiovasc Imaging 2009;25:45562; with permission.).

508

Garcia & Movassaghi

Fig. 5. The development of new imaging trajectories. The left coronary artery dual-motion rotational trajectory is shown in yellow on each individual vessel segment average optimal view map. The green lines show the two in plane rotational acquisitions. The red regions show the 10% average foreshortening and the white dot gives the single best optimal view along the trajectory. The left coronary artery dual-axis trajectory allows optimal visualization of all segments and includes most major fixed views traditionally recommended. Caud, caudal; Cran, cranial; LAD, left anterior descending artery; LAO, left anterior oblique; LCX, left circumflex; OM1, first obtuse marginal; RAO, right anterior oblique.

through statistical characterization of this data can provide objective and quantitative characterization of vascular anatomy serving different purposes.

Cataloging Three-dimensional Vascular Features

Databases or catalogs of 3D anatomical features with subsequent statistical characterization of this human in vivo data are useful for a variety of applications. For example, catalogs of 3D anatomical features can assist in the design and testing of different cardiovascular devices. These types of data enable the interventionalist to place the anatomy of the patient being treated in the context of anatomies of similar patients, such as those with similar bifurcation angles, and then plan accordingly (Fig. 6). Furthermore this database can be used for developing correlative statistics to clinical outcomes of interventions. Traditional lesion classification systems have been using 2D angiographic images and have been shown to be useful in predicting the outcome of coronary interventions and the need to use adjunctive devices.13 Clinical features combined with these anatomical features can predict

complications. Anatomical features alone are particularly powerful in predicting success or failure. Validation studies have been completed of the predictive capabilities the American College of Cardiology (ACC)/American Heart Association (AHA) and the Society of Coronary Angiography and Interventions (SCAI) lesion classification systems, yet the limitations of using 2D angiographic data also have been defined with only fair interobserver agreement in lesion classification.13 This should not be surprising because lesion characteristics and success in treating lesions are heavily dependent on subjective aspects of human performance not only in classifying lesions but in acquiring suitable angiographic views to clearly define the anatomy.14 In addition, many of the major characteristics (lesion length, accessibility/tortuosity, angulation, and lesion eccentricity) may be misrepresented and poorly quantified in 2D projection images because of foreshortening, overlap, and other limitations of traditional angiography. As pointed out in the recently released percutaneous coronary intervention guidelines, no prospective studies using core laboratory analysis have validated systems of

Computer Assistance to Solve Imaging Tasks

509

Fig. 6. (A) Posterior descending artery (PDA) bifurcation angle database. Note the variation in angle behavior in these 100 right coronary arteryposterior descending artery bifurcations. The spectrum of angles goes from acute to obtuse. (B) Right coronary artery curvature database. This is the spectrum of radius of curvature in 100 right coronary arteries. The spectrum of curvature goes from 0 to 55 .

lesion and target-vessel classification to stratify the risk of success and complications. Alternatively more elaborate, sophisticated, objective, and standardized lesion classification

systems can be designed and tested for predictive value for interventional outcomes using 3D computer assistance. For example, tortuosity can be measured in units of curvature/torsion and

510

Garcia & Movassaghi

lengths in 3D space can be accurately measured. Finally, the use of these more advanced lesion classification systems using 3D vascular imaging may be extended to predict not simply success versus failure, but also to prospectively identify the long high-radiation dose and large contract volume cases and the need for certain specialized equipment, including navigation systems. pre-planning of procedures resulting in preprogrammed clinically meaningful image gantries.41 Ideally these gantries, by making the trial-anderror technique obsolete, will decreasing the use of contrast media and radiation exposure thus enhance patient safety. Current rotational acquisitions may soon provide data that can be automatically processed by the computer. This would eliminate user interaction and providing in-room real-time data that can potentially improve procedural outcomes. Additionally, recent advances in noninvasive coronary evaluation will deliver data that can be used for planning and, in certain circumstances, the fusion of imaging technologies. These computer advances have not only improved the in-room flow of coronary interventions but have also served as a platform for medical education through computer-based simulations.50 Data from all imaging technologies, collected in a database and coupled with medical simulation, can form the basis of a realistic environment for practicing procedures with similar characteristics.50 While x-raybased angiography is an old technology, its diagnostic and therapeutic usefulness continues to improve through advances in computer assistance.

Simulation of Cardiovascular Procedures

Another application for computer assistance and 3D vascular data is in simulation of medical procedures.50 A system for simulating a fluoroscopic medical procedure first must enable an operator to simulate the performance of a procedure in a vascular tree. The system should also, however, give the operator the option of selecting any gantry position. Furthermore, the 3D vascular trees are needed for the operator to master the hand-eye coordination tasks of the interventional technique being simulated in a realistic vascular system. Current medical procedure simulators already use patient-derived 3D vascular data in the simulation of coronary, carotid, and peripheral vascular interventions. Three-dimensional vascular trees derived from diagnostic CT and MR angiograms and downloaded into a procedure simulator will allow the operator to practice an intervention before actually performing it. To make the most of this novel application, 3D data files must be standardized and simulation technology must be refined to be able immediately incorporate patient-specific data into a simulated case structure. Further studies will be needed to evaluate the impact of simulation training in procedural outcomes.

REFERENCES
1. Green NE, Chen SY, Messenger JC, et al. Threedimensional vascular angiography. Curr Probl Cardiol 2004;29(3):10442. 2. Topol EJ, Nissen SE. Our preoccupation with coronary luminology. The dissociation between clinical and angiographic findings in ischemic heart disease. Circulation 1995;92(8):233342. 3. Agostoni P, Biondi-Zoccai G, Van Langenhove G, et al. Comparison of assessment of native coronary arteries by standard versus three-dimensional coronary angiography. Am J Cardiol 2008;102(3):2729. 4. Garcia JA. Optimal angiographic views based on 3D reconstructed models. J Am Coll Cardiol 2007; 49(Suppl B:9):296A. 5. Garcia JA. Image guidance of percutaneous coronary and structural heart disease interventions using a CT and fluoroscopy integration. Vascular Disease Management 2007;4(3):14. 6. Garcia JA, Bhakta S, Kay J, et al. On-line multi-slice computed tomography interactive overlay with conventional x-ray: a new and advanced imaging fusion concept. Int J Cardiol 2009;133(3):e1015. Epub Jan 29, 2008. 7. Garcia JA, Chen J, Hansgen A, et al. Rotational angiography (RA) and three-dimensional imaging

SUMMARY
Traditional coronary angiography continues to evolve with ongoing changes in technology. Current computer-based advancements have led to the transformation of cine-based acquisitions into all-digital formats. These all-digital formats allow for the advanced processing of acquired images and have paved the way for 3D modeling, 3D reconstructions, advanced computer graphics, and optimal view maps. Three-dimensional modeling has been well described and provides the basis for a fast inroom evaluation, which, coupled with optimal view maps, aims at minimizing imaging inaccuracies while providing a superior safety profile through less exposure to radiation and contrast.3,4,39,41 Universal and, most importantly, patientspecific optimal view maps will allow for the

Computer Assistance to Solve Imaging Tasks

(3-DRA): an available clinical tool. Int J Cardiovasc Imaging 2007;23(1):913. Chen SJ, Carroll JD. 3-D reconstruction of coronary arterial tree to optimize angiographic visualization. IEEE Trans Med Imaging 2000;19(4):31836. Chen SJ, Hoffmann KR, Carroll JD. Three-dimensional reconstruction of coronary arterial tree based on biplane angiograms. SPIE. Medical Imaging. 1996;2710:10314. Chen SY, Carroll JD. Kinematic and deformation analysis of 4-D coronary arterial trees reconstructed from cine angiograms. IEEE Trans Med Imaging 2003;22(6):71021. Chen SY, Carroll JD, Messenger JC. Quantitative analysis of reconstructed 3-D coronary arterial tree and intracoronary devices. IEEE Trans Med Imaging 2002;21(7):72440. Messenger JC, Chen SY, Carroll JD, et al. 3D coronary reconstruction from routine single-plane coronary angiograms: clinical validation and quantitative analysis of the right coronary artery in 100 patients. Int J Card Imaging 2000;16(6):41327. Scanlon PJ, Faxon DP, Audet AM, et al. ACC/AHA guidelines for coronary angiography. A report of the American College of Cardiology/American Heart Association Task Force on practice guidelines (Committee on Coronary Angiography). Developed in collaboration with the Society for Cardiac Angiography and Interventions. J Am Coll Cardiol 1999; 33(6):1756824. Green NE, Chen SY, Hansgen AR, et al. Angiographic views used for percutaneous coronary interventions: a three-dimensional analysis of physician-determined vs. computer-generated views. Catheter Cardiovasc Interv 2005;64(4):4519. Chen SY, Metz CE. Improved determination of biplane imaging geometry from two projection images and its application to three-dimensional reconstruction of coronary arterial trees. Med Phys 1997;24(5):63354. Ding Z, Friedman MH. Quantification of 3-D coronary arterial motion using clinical biplane cineangiograms. Int J Card Imaging 2000;16(5):33146. Gross MF, Friedman MH. Dynamics of coronary artery curvature obtained from biplane cineangiograms. J Biomech 1998;31(5):47984. Hoffmann KR, Metz CE, Chen Y. Determination of 3D imaging geometry and object configurations from two biplane views: an enhancement of the Metz-Fencil technique. Med Phys 1995;22(8): 121927. Movassaghi B, Rasche V, Grass M, et al. A quantitative analysis of 3-D coronary modeling from two or more projection images. IEEE Trans Med Imaging 2004;23(12):151731. Shechter G, Devernay F, Coste-Maniere E, et al. Three-dimensional motion tracking of coronary arteries in biplane cineangiograms. IEEE Trans Med Imaging 2003;22(4):493503. Becker A, Leber A, White CW, et al. Multislice computed tomography for determination of coronary artery disease in a symptomatic patient population. Int J Cardiovasc Imaging 2007;23(3):3617. Epub Dec 8, 2006. Budoff MJ. Noninvasive coronary angiography using computed tomography. Expert Rev Cardiovasc Ther 2005;3(1):12332. Budoff MJ, Gopal A, Gul KM, et al. Prevalence of obstructive coronary artery disease in an outpatient cardiac CT angiography environment. Int J Cardiol 2008;129(1):326. Budoff MJ, Rasouli ML, Shavelle DM, et al. Cardiac CT angiography (CTA) and nuclear myocardial perfusion imaging (MPI)a comparison in detecting significant coronary artery disease. Acad Radiol 2007;14(3):2527. de Feyter P, Mollet NR, Cadermartiri F, et al. MS-CT coronary imaging. J Interv Cardiol 2003;16(6):4658. Gaemperli O, Schepis T, Koepfli P, et al. Accuracy of 64-slice CT angiography for the detection of functionally relevant coronary stenoses as assessed with myocardial perfusion SPECT. Eur J Nucl Med Mol Imaging 2007;34(8):116271. Epub Jan 12, 2007. Johnson TR, Nikolaou K, Wintersperger BJ, et al. Dual-source CT cardiac imaging: initial experience. Eur Radiol 2006;16(7):140915. Epub May 13, 2006. Langerak SE, Vliegen HW, Jukema JW, et al. Value of magnetic resonance imaging for the noninvasive detection of stenosis in coronary artery bypass grafts and recipient coronary arteries. Circulation 2003;107(11):15028. Langreck H, Schnackenburg B, Nehrke K, et al. MR coronary artery imaging with 3D motion adapted gating (MAG) in comparison to a standard prospective navigator technique. J Cardiovasc Magn Reson 2005;7(5):7937. Leber AW, Becker A, Knez A, et al. Accuracy of 64slice computed tomography to classify and quantify plaque volumes in the proximal coronary system: a comparative study using intravascular ultrasound. J Am Coll Cardiol 2006;47(3):6727. Leber AW, Knez A, Becker A, et al. Visualising noncalcified coronary plaques by CT. Int J Cardiovasc Imaging 2005;21(1):5561. Leschka S, Alkadhi H, Plass A, et al. Accuracy of MSCT coronary angiography with 64-slice technology: first experience. Eur Heart J 2005;26(15): 14827. Manning WJ, Nezafat R, Appelbaum E, et al. Coronary magnetic resonance imaging. Magn Reson Imaging Clin N Am 2007;15(4):60937, vii. Nikolaou K, Knez A, Rist C, et al. Accuracy of 64MDCT in the diagnosis of ischemic heart disease. AJR Am J Roentgenol 2006;187(1):1117.

511

8.

21.

9.

22.

10.

23.

11.

24.

12.

25. 26.

13.

27.

28.

14.

29.

15.

30.

16.

17.

31.

18.

32.

19.

33.

34.

20.

512

Garcia & Movassaghi

35. Paulin S, von Schulthess GK, Fossel E, et al. MR imaging of the aortic root and proximal coronary arteries. AJR Am J Roentgenol 1987;148(4): 66570. 36. Sakuma H. [Cardiac MRI: current status and future prospective]. Nippon Igaku Hoshasen Gakkai Zasshi 2003;63(1):215 [in Japanese]. 37. Sakuma H [Cardiac magnetic resonance imaging]. Kyobu Geka 2007;60(8 Suppl):63541 [in Japanese]. 38. Sakuma H, Goto M, Nomura Y, et al. Three-dimensional coronary magnetic resonance angiography with injection of extracellular contrast medium. Invest Radiol 1999;34(8):5038. 39. Garcia JA, Chen J, Hansgen A, et al. Rotational angiography (RA) and three-dimensional imaging (3-DRA): an available clinical tool. Int J Cardiovasc Imaging 2007;23(1):913. Epub Jun 16, 2006. 40. Garcia JA, Chen SY, Messenger JC, et al. Initial clinical experience of selective coronary angiography using one prolonged injection and a 180 degrees rotational trajectory. Catheter Cardiovasc Interv 2007 Aug 1;70(2):1906. 41. Garcia JA, Movassaghi B, Casserly IP, et al. Determination of optimal viewing regions for x-ray coronary angiography based on a quantitative analysis of 3D reconstructed models. Int J Cardiovasc Imaging 2008 Dec 20 [Epub ahead of print]. 42. Gollapudi RR, Valencia R, Lee SS, et al. Utility of three-dimensional reconstruction of coronary angiography to guide percutaneous coronary intervention. Catheter Cardiovasc Interv 2007;69(4):47982. 43. Gradaus R, Mathies K, Breithardt G, et al. Clinical assessment of a new real time 3D quantitative coronary angiography system: evaluation in stented vessel segments. Catheter Cardiovasc Interv 2006; 68(1):449. 44. Movassaghi B, Schaefer D, Grass M, et al. 3D reconstruction of coronary stents in vivo based on motion compensated x-ray angiograms. Med Image Comput Comput Assist Interv Int Conf Med Image Comput Comput Assist Interv 2006;9(Pt 2):17784. 45. Rasche V, Buecker A, Grass M, et al. ECG-gated 3D-rotational coronary angiography (3DRCA). Paris (France): Springer; 2002. 46. Rasche V, Movassaghi B, Grass M, et al. Automatic selection of the optimal cardiac phase for gated three-dimensional coronary x-ray angiography. Acad Radiol 2006;13(5):63040. 47. Budoff MJ, Gopal A, Gopalakrishnan D. Cardiac computed tomography: diagnostic utility and integration in clinical practice. Clin Cardiol 2006; 29(9 Suppl 1):I4I14. 48. de Feyter PJ, Nieman K. New coronary imaging techniques: what to expect? Heart 2002;87(3):1957. 49. Hecht HS. Applications of multislice coronary computed tomographic angiography to percutaneous coronary intervention: How did we ever do without it? Catheter Cardiovasc Interv 2008;71(4): 490503. 50. Carroll JD, Messenger JC. Medical simulation: the new tool for training and skill assessment. Perspect Biol Med 2008;51(1):4760.